@misc{13543, abstract = {{Neue Sonderausstellung “A KInd of Art. Künstliche Intelligenz trifft (Weser-)Renaissance” im Weserrenaissance-Museum Schloss Brake Was haben das schillernde Zeitalter der (Weser-)Renaissance und der Bereich der künstlichen Intelligenz bloß miteinander zu tun? Erstaunlich viel! Das innovative Weserrenaissance-Museum Schloss Brake zeigt in seiner topaktuellen Sonderausstellung “A KInd of Art. Künstliche Intelligenz trifft (Weser-)Renaissance” überraschende Parallelen und faszinierende Zusammenhänge auf, die kaum jemand vermuten würde. “Zusammen mit der TH OWL und Fraunhofer IOSB-INA wagen wir den Sprung von der Zwei- in die Dreidimensionalität und zeigen Deutschlands erste Museumsausstellung mit KI-Skulpturen, die aus historischen Exponaten entwickelt wurden. Diese weisen allesamt einen unmittelbaren Bezug zur (Weser-)Renaissance auf und kombinieren die Vergangenheit und die Zukunft aufs Vortrefflichste miteinander”, sagt Museumsleiterin Silvia Herrmann. Die KI-Skulpturen stammen dabei allesamt aus einem Master-Kurs-Projekt des Fachbereiches Medienproduktion unter der Leitung von Prof. Anke Stache. “Hätten Sie beispielsweise gewusst, dass das weltberühmte Universalgenie Leonardo da Vinci bereits vor mehr als 500 Jahren einen Automaten entwickelt hat? Es ist uns gelungen, ein nachgebautes und bewegliches Modell seines ‘Roboter-Ritters’ als Leihgabe für die Ausstellung zu gewinnen”, sagt die Kuratorin Dr. Susanne Hilker. Passend dazu treffen Leonardo da Vinci und der Roboter Ina in Form eines Comics fiktiv aufeinander und unterhalten sich über die Innovationen ihrer jeweiligen Zeit. Zu bestaunen sind auch zahlreiche kunsthistorische Originale wie beispielsweise “Minerva und die Musen auf dem Helikon” von Hans Rottenhammer oder “Die Tempelreinigung” von Hans und Paul Vredeman de Vries. Auch hierbei gibt es spannende Verbindungen zum Gebiet der künstlichen Intelligenz. Freuen können sich die Besucher auch auf Mitmachstationen wie eine interaktive Klanginstallation, ein Ergometer, mit dessen Hilfe sie herausfinden können, wie viel Energie die künstliche Intelligenz für ihre Prozesse benötigt, und auf eine Fotobox. Apropos Fotos: Im Rahmen dieser Ausstellung stellt das Museum auch originelle Kunstdoppelgänger aus. Darüber hinaus sind im “Freiraum” die 20 besten KI-generierten Bilder zu sehen, die im Rahmen eines Wettbewerbs des CIIT (Centrum Industrial IT) entstanden sind. Schlussendlich wird den Besuchern zur Abstimmung die Frage aller Fragen gestellt: Kann KI Kunst? “Wir möchten mit dieser Sonderausstellung zeigen, wie innovativ, interessant und relevant Museen sein können. Ganz bewusst greifen wir dieses topaktuelle Thema auf. Wir möchten in puncto Künstliche Intelligenz zum Nachdenken und zur Diskussion anregen. Darüber hinaus schlagen wir eine Brücke zwischen Vergangenheit, Gegenwart und Zukunft und eröffnen neue Perspektiven”, sagt Silvia Herrmann. Passend zur neuen Sonderausstellung bietet das Weserrenaissance-Museum Schloss Brake kurzweilige Mitmachprogramme für Kindergärten und Schulen an. Des Weiteren stehen zahlreiche Veranstaltungen mit Bezug zum Thema KI auf dem Programm. Alle Infos unter www.museum-schloss-brake.de Und wer von künstlicher Intelligenz nicht genug bekommen kann, macht einen Ausflug zum Kooperationspartner, der “Eulenburg”. Das Universitäts- und Stadtmuseum Rinteln zeigt ebenfalls eine Ausstellung mit KI-generierten Skulpturen. Die Ausstellung wird gefördert vom Ministerium für Kultur und Wissenschaft des Landes NRW und dem Regionalen Kultur Programm NRW. Die Ausstellung findet in Kooperation mit folgenden Partnern statt: Fraunhofer IOSB-INA, Technische Hochschule OWL, Kl Akademie OWL, inIT TH OWL, Bundesministerium für Forschung, Technologie und Raumfahrt, KreativInstitut.OWL, Trinnovation OWL, Hochschule für Musik Detmold, Fachbereich Ingenieurwissenschaften und Mathematik der Hochschule Bielefeld HSBI, LWL Museum Ziegelei Lage und wird unterstützt von den „Frauen für Lemgo”. Das Weserrenaissance-Museum Schloss Brake dankt auch seinem Träger, dem Landesverband Lippe, den Mitfinanziers, dem LWL sowie der Alten Hansestadt Lemgo, sowie den Sponsoren, der Lippischen Landesbrandversicherung AG und der Sparkasse Lemgo, für die Unterstützung! Der Eintritt in die neue Sonderausstellung im Weserrenaissance-Museum Schloss Brake beträgt 7 Euro. Kinder und Jugendliche bis 18 Jahre haben freien Eintritt. Die Ausstellung kann zwischen dem 16. September 2025 und 01. Februar 2026 dienstags bis sonntags von 10 bis 18 Uhr besichtigt werden. (https://museum-schloss-brake.de/sonderausstellung/)}}, author = {{Lange-Hegermann, Markus}}, keywords = {{KI, Kunst, Weserrenaissance}}, publisher = {{Landesverband Lippe}}, title = {{{A KInd of Art}}}, year = {{2025}}, } @misc{11605, abstract = {{The recovery of beer from surplus yeast is to date an economical business case only for large breweries. In this work, a here novel process with rotating ceramic microfiltration membranes is used. This allows a very high lift force to be achieved while still maintaining a small transmembrane pressure to reduce the formation of a fouling layer. The results show that long running times (between cleanings) are possible, limited only by the change in the rheological properties of the suspension due to thickening. From a so-called "Inflexion Point" (IF), the filtration behavior changes abruptly. The aim of the work was therefore to use machine learning aided modeling to predict the IF from experimental data in order to optimize the process and to achieve the most economical conditions. The economic efficiency depends on the space-time yields. The results show that a significant improvement in economic efficiency could be possible with the help of modeling and this special kind of filtration technology. However, the economic efficiency depends finally on the conditions in each individual brewery.}}, author = {{Trilling-Haasler, Marc and Tebbe, Jörn and Lange-Hegermann, Markus and Schneider, Jan}}, keywords = {{surplus yeast, membrane filtration, microfiltration}}, location = {{Lille}}, title = {{{Yeast filtration with rotating membrane filtration – a new approach for an economical recovery of beer form surplus yeast }}}, year = {{2024}}, } @misc{12815, abstract = {{Active learning of physical systems must commonly respect practical safety constraints, which restricts the exploration of the design space. Gaussian Processes (GPs) and their calibrated uncertainty estimations are widely used for this purpose. In many technical applications the design space is explored via continuous trajectories, along which the safety needs to be assessed. This is particularly challenging for strict safety requirements in GP methods, as it employs computationally expensive Monte-Carlo sampling of high quantiles. We address these challenges by providing provable safety bounds based on the adaptively sampled median of the supremum of the posterior GP. Our method significantly reduces the number of samples required for estimating high safety probabilities, resulting in faster evaluation without sacrificing accuracy and exploration speed. The effectiveness of our safe active learning approach is demonstrated through extensive simulations and validated using a real-world engine example.}}, author = {{Tebbe, Jörn and Zimmer, Christoph and Steland, Ansgar and Lange-Hegermann, Markus and Mies, Fabian}}, booktitle = {{International Conference on Artificial Intelligence and Statistics (AISTATS), Vol. 238}}, issn = {{2640-3498}}, location = {{Valencia, SPAIN}}, pages = {{1333--1341}}, publisher = {{MLResearchPress }}, title = {{{Efficiently Computable Safety Bounds for Gaussian Processes in Active Learning}}}, year = {{2024}}, } @misc{12816, abstract = {{Medical images need annotations with high-level semantic descriptors, so that domain experts can search for the desired dataset among an enormous volume of visual media within a Medical Data Integration Center. This article introduces a processing pipeline for storing and annotating DICOM and PNG imaging data by applying Elasticsearch, S3 and Deep Learning technologies. The proposed method processes both DICOM and PNG images to generate annotations. These image annotations are indexed in Elasticsearch with the corresponding raw data paths, where they can be retrieved and analyzed.}}, author = {{Cheng, Ka Yung and Pazmino, Santiago and Bergh, Bjoern and Lange-Hegermann, Markus and Schreiweis, Bjorn}}, booktitle = {{19th World Congress on Medical and Health Informatics (MEDINFO)}}, isbn = {{978-1-64368-456-7}}, issn = {{1879-8365}}, keywords = {{Medical image retrieval, data lake, DICOM, deep learning, elasticsearch}}, location = {{Sydney, AUSTRALIA}}, pages = {{1388--1389}}, publisher = {{IOS Press, Incorporated}}, title = {{{An Image Retrieval Pipeline in a Medical Data Integration Center.}}}, doi = {{10.3233/SHTI231208}}, volume = {{310}}, year = {{2024}}, } @misc{12822, abstract = {{A medical data integration center integrates a large volume of medical images from clinical departments, including X-rays, CT scans, and MRI scans. Ideally, all images should be indexed appropriately with standard clinical terms. However, some images have incorrect or missing annotations, which creates challenges in searching and integrating data centrally. To address this issue, accurate and meaningful descriptors are needed for indexing fields, enabling users to efficiently search for desired images and integrate them with international standards. This paper aims to provide concise annotation for missing or incorrectly indexed fields, incorporating essential instance -level information such as radiology modalities (e.g., X-rays), anatomical regions (e.g., chest), and body orientations (e.g., lateral) using a Deep Learning classification model - ResNet50. To demonstrate the capabilities of our algorithm in generating annotations for indexing fields, we conducted three experiments using two opensource datasets, the ROCO dataset, and the IRMA dataset, along with a custom dataset featuring SNOMED CT labels. While the outcomes of these experiments are satisfactory (Precision of >75%) for less critical tasks and serve as a valuable testing ground for image retrieval, they also underscore the need for further exploration of potential challenges. This essay elaborates on the identified issues and presents well-founded recommendations for refining and advancing our proposed approach.}}, author = {{Cheng, Ka Yung and Lange-Hegermann, Markus and Hövener, Jan-Bernd and Schreiweis, Björn}}, booktitle = {{Computational and Structural Biotechnology Journal}}, issn = {{2001-0370}}, keywords = {{DICOM images, Medical image captioning, Medical image interchange, SNOMED CT body structure}}, pages = {{434--450}}, publisher = {{Elsevier BV}}, title = {{{Instance-level medical image classification for text-based retrieval in a medical data integration center}}}, doi = {{10.1016/j.csbj.2024.06.006}}, volume = {{24}}, year = {{2024}}, } @misc{11381, author = {{Hernández Rodriguez, Tanja and Ramm, Selina and Lange-Hegermann, Markus and Frahm, Björn}}, location = {{Berlin, Germany}}, publisher = {{DECHEMA e.V.}}, title = {{{A systematic, model-based workflow for risk-based decision making in upstream development}}}, year = {{2023}}, } @misc{11383, author = {{Hernández Rodriguez, Tanja and Posch, Christoph and Pörtner, Ralf and Lange-Hegermann, Markus and Wurm, Florian M. and Frahm, Björn}}, location = {{Berlin, Germany}}, publisher = {{DECHEMA e.V.}}, title = {{{Model-assisted design strategies for bioprocesses – Advanced statistical methods in industrial upstream cell culture}}}, year = {{2023}}, } @misc{10201, author = {{Hernández Rodriguez, Tanja and Ramm, Selina and Lange-Hegermann, Markus and Frahm, Björn}}, location = {{Recklinghausen, Germany}}, title = {{{A systematic, model-based workflow for risk-based decision making in upstream development}}}, year = {{2023}}, } @misc{10787, abstract = {{Cyber-physical production systems have emerged with the rise of Industry 4.0 in different industrial fields. Especially the food sector, where inhomogeneous input products like beer/yeast suspensions with different qualities and properties have yet slowed down automation, has potential for this evolution. This contribution presents optimization methods for a dynamical cross-flow filtration plant which is driven by an advanced control concept in combination with data driven product monitoring via inline near infrared spectroscopy (NIR) in order to improve energy savings and filtration performance. Using a hierarchical control and optimization structure, the non stationary batch process is steered towards a high production rate with low energy consumption for a variety of different input products.}}, author = {{Tebbe, Jörn and Pawlik, Thomas and Trilling-Haasler, Marc and Löbner, Jannis and Lange-Hegermann, Markus and Schneider, Jan}}, booktitle = {{2023 IEEE 21st International Conference on Industrial Informatics (INDIN)}}, editor = {{Jasperneite, Jürgen and Wisniewski, Lukasz and Fung Man, Kim}}, isbn = {{978-1-6654-9314-7 }}, issn = {{1935-4576}}, keywords = {{Spectroscopy, Production systems, Filtration, Velocity control, Optimization methods, Cyber-physical systems, Nonhomogeneous media}}, location = {{Lemgo}}, pages = {{1--7}}, publisher = {{IEEE}}, title = {{{Holistic optimization of a dynamic cross-flow filtration process towards a cyber-physical system}}}, doi = {{10.1109/INDIN51400.2023.10217913}}, year = {{2023}}, } @misc{12811, abstract = {{For environmental, sustainable economic and political reasons, recycling processes are becoming increasingly important, aiming at a much higher use of secondary raw materials. Currently, for the copper and aluminum industries, no method for the non-destructive online analysis of heterogeneous materials is available. The prompt gamma neutron activation analysis (PGNAA) has the potential to overcome this challenge. A difficulty when using PGNAA for online classification arises from the small amount of noisy data, due to short-term measurements. In this case, classical evaluation methods using detailed peak by peak analysis fail. Therefore, we propose to view spectral data as probability distributions. Then, we can classify material using maximum log-likelihood with respect to kernel density estimation and use discrete sampling to optimize hyperparameters. For measurements of pure aluminum alloys we achieve near-perfect classification of aluminum alloys under 0.25 s.}}, author = {{Shayan, Helmand and Krycki, Kai and Doemeland, Marco and Lange-Hegermann, Markus}}, booktitle = {{IEEE Transactions on Nuclear Science}}, issn = {{1558-1578}}, keywords = {{Classification of metal, kernel density estimation, maximum log-likelihood, online classification, prompt gamma neutron activation analysis (PGNAA) spectral classification, random sampling}}, number = {{6}}, pages = {{1171--1177}}, publisher = {{IEEE}}, title = {{{PGNAA Spectral Classification of Metal With Density Estimations}}}, doi = {{10.1109/tns.2023.3242626}}, volume = {{70}}, year = {{2023}}, } @misc{12828, abstract = {{Partial differential equations (PDEs) are important tools to model physical systems and including them into machine learning models is an important way of incorporating physical knowledge. Given any system of linear PDEs with constant coefficients, we propose a family of Gaussian process (GP) priors, which we call EPGP, such that all realizations are exact solutions of this system. We apply the Ehrenpreis-Palamodov fundamental principle, which works as a non-linear Fourier transform, to construct GP kernels mirroring standard spectral methods for GPs. Our approach can infer probable solutions of linear PDE systems from any data such as noisy measurements, or pointwise defined initial and boundary conditions. Constructing EPGP-priors is algorithmic, generally applicable, and comes with a sparse version (S-EPGP) that learns the relevant spectral frequencies and works better for big data sets. We demonstrate our approach on three families of systems of PDEs, the heat equation, wave equation, and Maxwell's equations, where we improve upon the state of the art in computation time and precision, in some experiments by several orders of magnitude.}}, author = {{Härkönen, Marc and Lange-Hegermann, Markus and Raiţă, Bogdan}}, booktitle = {{40th International Conference on Machine Learning}}, issn = {{2640-3498}}, location = {{Honolulu, HI}}, publisher = {{MLResearchPress }}, title = {{{Gaussian Process Priors for Systems of Linear Partial Differential Equations with Constant Coefficients}}}, volume = {{202}}, year = {{2023}}, } @misc{9930, author = {{Lange-Hegermann, Markus and Schmohl, Tobias and Watanabe, Alice and Schelling, Kathrin and Heiss, Stefan and Rubart, Jessica}}, booktitle = {{Künstliche Intelligenz in der Hochschulbildung: Chancen und Grenzen des KI-gestützten Lernens und Lehrens}}, editor = {{Schmohl, Tobias and Watanabe, Alice and Schelling, Kathrin}}, isbn = {{978-3-8376-5769-2}}, pages = {{161--172}}, publisher = {{transcript Verlag}}, title = {{{KI-basierte Erstellung individualisierter Mathematikaufgaben für MINT-Fächer}}}, doi = {{10.14361/9783839457696-009}}, volume = {{4}}, year = {{2023}}, } @misc{11377, abstract = {{consuming and often performed rather empirically. Efficient optimization of multiple objectives such as process time, viable cell density, number of operating steps & cultivation scales, required medium, amount of product as well as product quality depicts a promising approach. This contribution presents a workflow which couples uncertainty-based upstream simulation and Bayes optimization using Gaussian processes. Its application is demonstrated in a simulation case study for a relevant industrial task in process development, the design of a robust cell culture expansion process (seed train), meaning that despite uncertainties and variabilities concerning cell growth, low variations of viable cell density during the seed train are obtained. Compared to a non-optimized reference seed train, the optimized process showed much lower deviation rates regarding viable cell densities (<10% instead of 41.7%) using five or four shake flask scales and seed train duration could be reduced by 56 h from 576 h to 520 h. Overall, it is shown that applying Bayes optimization allows for optimization of a multi-objective optimization function with several optimizable input variables and under a considerable amount of constraints with a low computational effort. This approach provides the potential to be used in the form of a decision tool, e.g., for the choice of an optimal and robust seed train design or for further optimization tasks within process development.}}, author = {{Hernández Rodriguez, Tanja and Sekulic, Anton and Lange-Hegermann, Markus and Frahm, Björn}}, booktitle = {{Processes}}, issn = {{2227-9717}}, keywords = {{Gaussian processes, Bayes optimization, Pareto optimization, multi-objective, cell culture, seed train}}, number = {{5}}, publisher = {{MDPI AG}}, title = {{{Designing Robust Biotechnological Processes Regarding Variabilities Using Multi-Objective Optimization Applied to a Biopharmaceutical Seed Train Design}}}, doi = {{10.3390/pr10050883}}, volume = {{10}}, year = {{2022}}, } @misc{7932, author = {{Hernández Rodriguez, Tanja and Ramm, Selina and Lange-Hegermann, Markus and Frahm, Björn}}, location = {{Barcelona, Spain}}, title = {{{A systematic, model-based workflow for risk-based decision making in upstream development}}}, year = {{2022}}, } @inbook{10193, abstract = {{Development and optimization of biopharmaceutical production processes with cell cultures is cost- and time-consuming and often performed rather empirically. Efficient optimization of multiple objectives such as process time, viable cell density, number of operating steps & cultivation scales, required medium, amount of product as well as product quality depicts a promising approach. This contribution presents a workflow which couples uncertainty-based upstream simulation and Bayes optimization using Gaussian processes. Its application is demonstrated in a simulation case study for a relevant industrial task in process development, the design of a robust cell culture expansion process (seed train), meaning that despite uncertainties and variabilities concerning cell growth, low variations of viable cell density during the seed train are obtained. Compared to a non-optimized reference seed train, the optimized process showed much lower deviation rates regarding viable cell densities (<10% instead of 41.7%) using five or four shake flask scales and seed train duration could be reduced by 56 h from 576 h to 520 h. Overall, it is shown that applying Bayes optimization allows for optimization of a multi-objective optimization function with several optimizable input variables and under a considerable amount of constraints with a low computational effort. This approach provides the potential to be used in the form of a decision tool, e.g., for the choice of an optimal and robust seed train design or for further optimization tasks within process development.}}, author = {{Hernández Rodriguez, Tanja and Sekulic, Anton and Lange-Hegermann, Markus and Frahm, Björn}}, booktitle = {{Bioprocess Systems Engineering Applications in Pharmaceutical Manufacturing}}, editor = {{Pörtner, Ralf and Möller, Johannes}}, isbn = {{978-3-0365-5210-1}}, issn = {{2227-9717}}, keywords = {{Gaussian processes, Bayes optimization, Pareto optimization, multi-objective, cell culture, seed train}}, pages = {{21--48}}, publisher = {{MDPI}}, title = {{{Designing robust biotechnological processes regarding variabilities using multi-objective optimization applied to a biopharmaceutical seed train design}}}, doi = {{https://doi.org/10.3390/pr10050883}}, volume = {{special issue}}, year = {{2022}}, } @misc{10198, author = {{Hernández Rodriguez, Tanja and Pörtner, Ralf and Lange-Hegermann, Markus and Wurm, Florian M. and Frahm, Björn}}, location = {{Lisbon, Portugal }}, title = {{{A systematic, model-based approach for decision making in upstream development – Considerations regarding clone selection and cell expansion}}}, year = {{2022}}, } @misc{12804, abstract = {{Data in many applications follows systems of Ordinary Differential Equations (ODEs). This paper presents a novel algorithmic and symbolic construction for covariance functions of Gaussian Processes (GPs) with realizations strictly following a system of linear homogeneous ODEs with constant coefficients, which we call LODE-GPs. Introducing this strong inductive bias into a GP improves modelling of such data. Using smith normal form algorithms, a symbolic technique, we overcome two current restrictions in the state of the art: (1) the need for certain uniqueness conditions in the set of solutions, typically assumed in classical ODE solvers and their probabilistic counterparts, and (2) the restriction to controllable systems, typically assumed when encoding differential equations in covariance functions. We show the effectiveness of LODE-GPs in a number of experiments, for example learning physically interpretable parameters by maximizing the likelihood.}}, author = {{Besginow, Andreas and Lange-Hegermann, Markus}}, booktitle = {{36th Conference on Neural Information Processing Systems (NeurIPS 2022) }}, editor = {{Koyejo, S. and Mohamed, S. and Agarwal, A. and Belgrave, D. and Cho, K. and Oh, A.}}, isbn = {{978-1-7138-7108-8 }}, issn = {{1049-5258}}, keywords = {{SMITH NORMAL-FORM, ALGORITHMS, REDUCTION}}, location = {{New Orleans, La.; Online}}, pages = {{29386 -- 29399}}, publisher = {{Curran Associates, Inc.}}, title = {{{Constraining Gaussian Processes to Systems of Linear Ordinary Differential Equations}}}, volume = {{35}}, year = {{2022}}, } @misc{7581, author = {{Lange-Hegermann, Markus and Schmohl, Tobias and Watanabe, Alice and Heiss, Stefan and Rubart, Jessica}}, booktitle = {{New Perspectives in Science Education}}, location = {{Florenz}}, pages = {{385--390}}, publisher = {{Libreriauniversitaria.it}}, title = {{{AI-based STEM education: Generating individualized exercises in mathematics}}}, volume = {{10}}, year = {{2021}}, } @misc{5620, author = {{Lange-Hegermann, Markus and Schmohl, Tobias and Watanabe, Alice and Heiss, Stefan and Rubart, Jessica}}, booktitle = {{New Perspectives in Science Education}}, isbn = {{979-12-80225-14-6}}, location = {{Virtual Event}}, pages = {{385 -- 390}}, publisher = {{Libreriauniversitaria.it}}, title = {{{AI-Based Stem Education: Generating Individualized Exercises in Mathematics}}}, doi = {{10.26352/F318_2384-9509}}, year = {{2021}}, } @misc{12786, abstract = {{One goal in Bayesian machine learning is to encode prior knowledge into prior distributions, to model data efficiently. We consider prior knowledge from systems of linear partial differential equations together with their boundary conditions. We construct multi-output Gaussian process priors with realizations in the solution set of such systems, in particular only such solutions can be represented by Gaussian process regression. The construction is fully algorithmic via Grobner bases and it does not employ any approximation. It builds these priors combining two parametrizations via a pullback: the first parametrizes the solutions for the system of differential equations and the second parametrizes all functions adhering to the boundary conditions.}}, author = {{Lange-Hegermann, Markus}}, booktitle = {{24th International Conference on Artificial Intelligence and Statistics (AISTATS)}}, editor = {{Banerjee, A. and Fukumizu, K.}}, issn = {{2640-3498}}, keywords = {{FUNCTIONAL REGRESSION, PREDICTION, ALGORITHMS, COMPLEXITY, MODELS}}, location = {{Virtual}}, publisher = {{MLResearchPress }}, title = {{{Linearly Constrained Gaussian Processes with Boundary Conditions}}}, volume = {{130}}, year = {{2021}}, } @misc{12812, abstract = {{Discerning unexpected from expected data patterns is the key challenge of anomaly detection. Although a multitude of solutions has been applied to this modern Industry 4.0 problem, it remains an open research issue to identify the key characteristics subjacent to an anomaly, sc. generate hypothesis as to why they appear. In recent years, machine learning models have been regarded as universal solution for a wide range of problems. While most of them suffer from non-self-explanatory representations, Gaussian Processes (GPs) deliver interpretable and robust statistical data models, which are able to cope with unreliable, noisy, or partially missing data. Thus, we regard them as a suitable solution for detecting and appropriately representing anomalies and their respective characteristics. In this position paper, we discuss the problem of automatic and interpretable anomaly detection by means of GPs. That is, we elaborate on why GPs are well suited for anomaly detection and what the current challenges are when applying these probabilistic models to large-scale production data.}}, author = {{Berns, Fabian and Lange-Hegermann, Markus and Beecks, Christian}}, booktitle = {{ Proceedings of the International Conference on Innovative Intelligent Industrial Production and Logistics IN4PL - Volume 1}}, editor = {{Panetto, H. and Madani, K. and Smirnov, A.}}, isbn = {{978-989-758-476-3}}, keywords = {{Anomaly Detection, Gaussian Processes, Explainable Machine Learning, Industry 4.0}}, location = {{Budapest, HUNGARY}}, pages = {{87--92}}, publisher = {{SCITEPRESS - Science and Technology Publications}}, title = {{{Towards Gaussian Processes for Automatic and Interpretable Anomaly Detection in Industry 4.0}}}, doi = {{10.5220/0010130300870092}}, year = {{2020}}, }